linux-kernel - Re: [PATCH] x86/cpu/AMD: Fix CPB bit for more processors

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Fri, 16 Nov 2018 10:54:24 +0100
From:   Borislav Petkov <bp@...en8.de>
To:     Jiaxun Yang <jiaxun.yang@...goat.com>, sherry.hurwitz@....com
Cc:     x86@...nel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] x86/cpu/AMD: Fix CPB bit for more processors

On Fri, Nov 16, 2018 at 01:31:06PM +0800, Jiaxun Yang wrote:
> CPUID Fn8000_0007_EDX[CPB] is wrongly 0 on Model 17,
> Stepping 0, but revision guide has not been released for
> newer Family 17h models.
> 
> Tesed on AMD "Ryzen 7 2700U with Radeon Vega Mobile Gfx"
> and "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
> their CPUID Fn0000_0001_EAX is 0x00810f10 and should have
> CPB feature according AMD product specifications, however
> their Fn8000_0007_EDX is 0x00006599, indicating they don't
> support CPB feature.
> 
> Signed-off-by: Jiaxun Yang <jiaxun.yang@...goat.com>
> ---
>  arch/x86/kernel/cpu/amd.c | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
> index eeea634bee0a..7db43ef8e97e 100644
> --- a/arch/x86/kernel/cpu/amd.c
> +++ b/arch/x86/kernel/cpu/amd.c
> @@ -821,8 +821,12 @@ static void init_amd_zn(struct cpuinfo_x86 *c)
>  	/*
>  	 * Fix erratum 1076: CPB feature bit not being set in CPUID. It affects
>  	 * all up to and including B1.
> +	 *
> +	 * Revision guide for Family 17h, Model 17 has not been released, but
> +	 * Model 17, Stepping 0 have the same issue.
>  	 */
> -	if (c->x86_model <= 1 && c->x86_stepping <= 1)
> +	if ((c->x86_model <= 1 && c->x86_stepping <= 1) ||	\
> +		(c->x86_model == 17 && c->x86_stepping == 0))
>  		set_cpu_cap(c, X86_FEATURE_CPB);
>  }

Actually, I'd prefer if we stopped doing all the checks and simply do
this:

	if (!cpu_has(c, X86_FEATURE_CPB))
		set_cpu_cap(c, X86_FEATURE_CPB);

for all F17h as it is a safe assumption that family 0x17 has CPB -
regardless of CPUID bit setting.

Sherry?

-- 
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.